Maker Pro
Maker Pro

Frequency divider (less than 1MHz)

ehsan66

Mar 3, 2016
6
Joined
Mar 3, 2016
Messages
6
I am looking to buy/biuld a frequency divider (division factor 4) that has the following characteristics:

  • Input signal: Sine wave- frequency in range of less than 1MHz (usually in range of 100-300 kOhm) input level >14dBm
  • Output signal: Sine wave- Amplitude should have a level of around 14dBm (be in order of 3 to 5 Vrms. Resistance around 1.1 to 1.5 KOhm)
I have been looking for a commercial product, but it seems that most of frequency dividers work in a MHz range and not in KHz range. Any ideas?
 
Last edited by a moderator:

davenn

Moderator
Sep 5, 2009
14,085
Joined
Sep 5, 2009
Messages
14,085
hi there
welcome to EP :)

not sure you can do that on a sine wave

the MHz prescaler IC's, ( dividers) eg MC12079,generally produce a square wave out for feeding to a PLL IC ( Phase Locked Loop/synthesiser)

the usual way to a lower AC freq from a higher one is by using a mixer
where you mix the incoming signal with a local oscillator ( LO) signal to produce a resulting output

3 outputs are actually produced ...
a strongly attenuated Original Frequency Fo
a lower freq Fo - LO
a higher freq Fo = LO

filtering is used to get rid of the 2 signals you don't want and leave the wanted one ... the high side one or the low side one


Dave
 
Last edited:

hevans1944

Hop - AC8NS
Jun 21, 2012
4,762
Joined
Jun 21, 2012
Messages
4,762
Getting the input sine wave frequency divided by four is the easy part. You need a Schmitt trigger to convert the sine wave to square pulses which you then supply as clock signals to a divide-by-four binary counter. The counter output will be square waves at one fourth the frequency of the sine wave input. Now here comes the hard part: if you want and need a sine wave output, you are going to need a lot more circuitry.

Please tell us what you are trying to do, instead of telling us what you think you need: a "black box" that takes sine waves for input and produces sine waves at one fourth the frequency for output. That "end use" information will elicit much better responses here.
 

davenn

Moderator
Sep 5, 2009
14,085
Joined
Sep 5, 2009
Messages
14,085
Now here comes the hard part: if you want and need a sine wave output, you are going to need a lot more circuitry.

since he stated the need for a sine output, that is why I didn't go that explanation route

to keep it a sine all the way through freq mixing is about the only way I'm aware of
cuz as you said any other way gets very complicated ;)


D
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
12,782
Joined
Nov 17, 2011
Messages
12,782
to keep it a sine all the way through freq mixing is about the only way I'm aware of
Right, but thats additive, not multiplicative.

How about a VCO (sine output) which is controlled via a PLL loop? The input sine and the output sine need to be quantized e.g. by a Schmitt-trigger (post #3), the a /4 divider, phase comparator and analog loop filter generate the control voltage for the VCO. A 4046B chip can be used for teh PLL part. All that is required is an additional VCO with sine output as the 4046 built-in VCO generates a square wave signal.

This project could be a starting point.
 

davenn

Moderator
Sep 5, 2009
14,085
Joined
Sep 5, 2009
Messages
14,085
Right, but thats additive, not multiplicative.

.

doesn't need to be multiplicative
As I exampled it is additive or subtractive

The important part is it gives the required freq out and it's still a sine wave and you don't have to screw around with converting square waves back to sine waves
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
12,782
Joined
Nov 17, 2011
Messages
12,782
doesn't need to be multiplicative
As I exampled it is additive or subtractive

The important part is it gives the required freq out and it's still a sine wave and you don't have to screw around with converting square waves back to sine waves
But the op wants a multiplicative change in frequency by 1/4.
While subtracting a fixed frequency will result in 1/4 the input frequency for a single frequency, this will no longer be valid when the input frequency changes. But if the input frequency were fixed, there wouldn't be any need to down-convert it: one could have an oscillator with a frequency fixed at 1/4 of the nominal input frequency and start/stop it depending on the presence of the input signal.
 

dorke

Jun 20, 2015
2,342
Joined
Jun 20, 2015
Messages
2,342
Getting the input sine wave frequency divided by four is the easy part. You need a Schmitt trigger to convert the sine wave to square pulses which you then supply as clock signals to a divide-by-four binary counter. The counter output will be square waves at one fourth the frequency of the sine wave input. Now here comes the hard part: if you want and need a sine wave output, you are going to need a lot more circuitry.
Here is another (relatively simple?) idea:
Convert the above square wave to a triangular-one,
then convert the triangle-wave to sine-wave with this differential-pair circuit(taken from TI AN-263).

tri to sine.jpg
 

davenn

Moderator
Sep 5, 2009
14,085
Joined
Sep 5, 2009
Messages
14,085
But the op wants a multiplicative change in frequency by 1/4.
While subtracting a fixed frequency will result in 1/4 the input frequency for a single frequency, this will no longer be valid when the input frequency changes. But if the input frequency were fixed, there wouldn't be any need to down-convert it: one could have an oscillator with a frequency fixed at 1/4 of the nominal input frequency and start/stop it depending on the presence of the input signal.

fair comment

since the OP still hasn't responded to requests for further info on what he is specifically doing, we have no idea in if the input freq is fixed or wildly variable

lets everyone take a breather till @ehsan66 responds with more info :)


Dave
 

ehsan66

Mar 3, 2016
6
Joined
Mar 3, 2016
Messages
6
Thank you so much, all for your support. Let me clarify and be more specific on the source of the signal and what I need:

Two Direct Digital Synthesizers (DDS) are used to generate two waveform at different frequencies (one at 150 and one at 160 kHz, phase aligned ). These digital signals are summed and are output on one 16-bit, 10MHz Digital to Analog Converts (DAC). Now, I do not have access to the DDS and DAC because of warranty issues. I only have an analog output from the device which has two frequency contents one at let's say 150 kHz and one at 160 kHz. This is the given part of the problem and cannot be changed (I call it input signal now). For my application, The signal I need should be sinusoidal and have two frequency contents one at 150/4=37.5 kHz and one at 40 kHz, however, it is very important that the phase of this signal is locked to the phase of input signal.

The DDS is controlled by a PID loop and it changes the frequency of dual frequency signal, but changes are not very rapid (at most 10 kHz change).


I am sorry if my electronic language is poor. hope this makes it clear enough.
 
Last edited:

davenn

Moderator
Sep 5, 2009
14,085
Joined
Sep 5, 2009
Messages
14,085
OK
that's substantially more complex than what you first stated
personally I cant help with that one ... beyond my knowledge

lets see what the others have to say
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
12,782
Joined
Nov 17, 2011
Messages
12,782
I'm sorry, but this is very different from the original question. I know of no method to divide a compound signal. Donw-converting the signal with amixer as suggested by Dave (post #4) is the only way Ican imagine, but you will achieve fout=fin-fmix which is 1/4 fin only for fmix=3/4*fin. For all other frequencies he equation turns out fout <> 1/4*fin.#
The easiest way would to run the DDS at 1/4 of the current freqeuncy which will output the desired frequency without further modification.
 

BobK

Jan 5, 2010
7,682
Joined
Jan 5, 2010
Messages
7,682
A /D, FFT, DDS.

In words: Convert to a digital signal, analyze the frequencies and phases using Fast Fourier Transform, divide the frequencies by 4, then use Direct Digital Synthesis to create the divided frequencies.

I doubt that there is any simpler solution that would be practical.

Bob
 

dorke

Jun 20, 2015
2,342
Joined
Jun 20, 2015
Messages
2,342
Here is another,more complicated idea : "pitch shifting".

1. Sample the signal at a high frequency fs (more than twice the Nyquist rate-in your case 1M*2= 2Mhz) by a fast A/D with enough resolution.
2. For every 4 Samples ,
produce a single one "output sample" Vs.
3. Convert the Vs samples back to Anaolg with a D/A at fs/4 rate.
4. Filter the signal with an analog LPF of fs/8
5. If you have access to the DDS clock you can produce fs from it.

The result should be that all frequencies in the spectrum of the signal would be shifted to 1/4 of the original frequency.
 

hevans1944

Hop - AC8NS
Jun 21, 2012
4,762
Joined
Jun 21, 2012
Messages
4,762
Thank you so much, all for your support. Let me clarify and be more specific on the source of the signal and what I need:
Your subsequent comments didn't "clarify" anything for me.

Two Direct Digital Synthesizers (DDS) are used to generate two waveform at different frequencies (one at 150 and one at 160 kHz, phase aligned ). These digital signals are summed and are output on one 16-bit, 10MHz Digital to Analog Converts (DAC).
What does it mean that "two different frequencies" are phase aligned? What is the digital nature of the summed signals? Is this a digital summation of two digital signals using a 16-bit adder? Do the digital signals represent sine waves? Are both synthesized waveforms represented in the DAC output as an analog waveform? Is this waveform low-pass filtered to eliminate the DAC frequency components greater than the highest input frequency?

Now, I do not have access to the DDS and DAC because of warranty issues. I only have an analog output from the device which has two frequency contents one at let's say 150 kHz and one at 160 kHz.
Do you intend to process the two frequency components of the output as independent signals? The output of the DAC is NOT a sinusoid unless the two frequencies do NOT occur simultaneously, as in frequency-shift modulation, and then only after a delay between shifts in frequency does the output become sinusoidal again.

This is the given part of the problem and cannot be changed (I call it input signal now). For my application, The signal I need should be sinusoidal and have two frequency contents one at 150/4=37.5 kHz and one at 40 kHz, however, it is very important that the phase of this signal is locked to the phase of input signal.
A sinusoidal signal, by definition, has only ONE frequency component. The addition of two sinusoidal wave-forms of different frequencies does not produce a sinusoidal sum.

The DDS is controlled by a PID loop and it changes the frequency of dual frequency signal, but changes are not very rapid (at most 10 kHz change).
Is this "10 kHz change" the rate at which the PID loop adjusts the DDS, or is it the magnitude of the frequency adjustment?[/QUOTE]


I am sorry if my electronic language is poor. hope this makes it clear enough.
It isn't clear enough for me to understand what you are trying to DO. Please describe what you are trying to DO rather than specifying a "black box" solution you think will work. What is the device you are trying to interface with, whose internals cannot be touched "because of warranty issues"? Manufacturer and model number could be a big help. Do you have access to schematics and/or maintenance manuals? Is this a "one off" project or do you plan to make more than one?
 

ehsan66

Mar 3, 2016
6
Joined
Mar 3, 2016
Messages
6
I'm puzzled as to the "phase alignment". How can signals of 150kHz and 160kHz stay in phase with each other?
What I mean is that the 150 kHz and 150/4 kHz signals are phased locked while 160 and 160/4 kHZ are also phased locked separately.
Ehsan
 

ehsan66

Mar 3, 2016
6
Joined
Mar 3, 2016
Messages
6
Your subsequent comments didn't "clarify" anything for me.


What does it mean that "two different frequencies" are phase aligned? What is the digital nature of the summed signals? Is this a digital summation of two digital signals using a 16-bit adder? Do the digital signals represent sine waves? Are both synthesized waveforms represented in the DAC output as an analog waveform? Is this waveform low-pass filtered to eliminate the DAC frequency components greater than the highest input frequency?


Do you intend to process the two frequency components of the output as independent signals? The output of the DAC is NOT a sinusoid unless the two frequencies do NOT occur simultaneously, as in frequency-shift modulation, and then only after a delay between shifts in frequency does the output become sinusoidal again.


A sinusoidal signal, by definition, has only ONE frequency component. The addition of two sinusoidal wave-forms of different frequencies does not produce a sinusoidal sum.


Is this "10 kHz change" the rate at which the PID loop adjusts the DDS, or is it the magnitude of the frequency adjustment?





Thanks for the note.
I will try to be more specific here. I am working with an atomic force microscope and the signals I get are generated in the controller of the AFM (see the attached diagram of the ARC2, specifically the FPGA and DSP parts).
As far as I know, the controller has two DDS that can digitally generate sine waveform. BTW, I was wrong about phase alignment and please ignore that part. Output from two DDSs are summed and are available on one 16-bit, 10MHz DAC ( I don't know much about the adder). Then this signal is low-pass filtered.

I do not intend to work with two signals with two different frequencies, but I need one modulated signal that has two frequency components ( and you are absolutely right, the summation of two sine signals of different frequencies is not a sine signal, but a modulated one, my bad.). I don't have an exact number for rate of change of frequency (integral gain=500 Hz), but the absolute change of frequency contents is less than 10 kHZ.


About our application, I attached another diagram that should show what we need to do. Basically, we need to drive a afm probe with a modulated signal. Due to embedded physics , we expect that afm probe will have oscillations at the frequencies four times higher than contents of the drive signal (fourth harmonic response). The cantilever deflection signal is then sent to a dual lock-in to find the fourth harmonics. That is why I need the DDSs to generate the digital signals at 150 and 160 kHZ since they are used as reference signals of lock-in. By using the amplitude difference of these two harmonics as an error function, the frequency is being tracked as shown in the graph. I hope this is clear enough now. Please kindly ask questions if there is any unclear parts.

Thanks,
Ehsan
 

Attachments

  • afm.png
    afm.png
    342.3 KB · Views: 209
  • 2016-03-04 13.30.30.jpg
    2016-03-04 13.30.30.jpg
    222.8 KB · Views: 149
Last edited:

hevans1944

Hop - AC8NS
Jun 21, 2012
4,762
Joined
Jun 21, 2012
Messages
4,762
@ehsan66 Thank you for the detailed reply. The block diagram is appreciated, but I will have to print it out to examine it closely. Is there an on-line resource for this image? Also, are you also using an Asylum AFM or some other manufacturer? Please provide manufacturer and model number of the AFM to which you interface the ARC2 controller. BTW, that's a pretty nice "toy" you have there. I don't think I would tear into that either, while it is still in warranty!

What I "know" about AFM today is not based on personal experience with real equipment, but only what I "read in the papers" published on the subject. I did briefly follow all the excitement in the previous century, when the first scanning tunneling microscope was invented by IBM researchers Binnig and Rohrer. In the popular press of the time it was touted as being a revolutionary breakthrough, allowing scientists to see individual atoms for the first time. Yeah, riiight. But it was still quite remarkable and worth a Nobel prize in physics in 1986. AFM was demonstrated shortly after the STM, but it was based on cantilevered contact-force measurements instead of electron tunneling. AFAIK, AFM is still based on sensing the environment around a cantilevered probe by mechanical scanning of the target under the probe, whether that environmental sensing be mechanical forces, magnetic fields, electrical fields or whatever. The sensor and associated instrumentation may change, but the principle of precisely moving the target under a nearby cantilevered sensor has remained the same. There has been some improvement in the electronics since the 1980s.:D

I get the impression you are exciting the cantilevered beam with two frequencies in a non-contact measurement mode and somehow measuring the change in resonance as a function of sample position under the probe. Please elucidate how this is supposed to work, if that is indeed what you are doing.

No doubt I will have to do some on-line research and play catch-up to understand what you are trying to do. Are you also using an Asylum AFM? What AFM are you interfacing the ARC2 controller to?

Adding two sine waves together is not the same as mixing them, a non-linear multiplicative modulation process that yields sum and difference frequencies. For example, when two notes of different frequencies are sounded together in air, the ear will hear a "beat" frequency that is the difference in frequency between the notes. If the notes are close in frequency the beat is at a sub-audible frequency and is heard as amplitude modulation of the average frequency of the two tones. All this happens because the ear is a non-linear transducer. The original notes are still present, and with suitable instrumentation can be separated.

This appears to be what is going on with your two lock-in amplifiers separating the fourth harmonic components of the f1 /4 and f2 /4 excitation applied to the cantilever, which for some reason you haven't explained responds at the fourth harmonic of the excitation. I will have to take your word for that because I don't see two digital lock-in amplifiers implemented on the ARC2 block diagram, which is probably generic. The literature does say two lock-in amplifiers are available in the FPGA.

As for solving your problem to create f1 /4 and f2 /4 from the summation of f1 and f2... that might be a lot easier if f1 and f2 were independently available and each was used to control two phase-locked sinusoidal signal generators operating at one fourth the reference frequencies. The outputs of the two generators would then be summed to drive the cantilever beam. Are f1 and f2 independently available, perhaps on BNC connectors?

I did half of that circuitry once in the 1970s using TTL discrete logic and a PLL integrated circuit to control a waveform generator that produced sine waves from triangle waves to generate sinusoidal audio tones for a psychology experiment that needed precise tonal values that could be digitally selected. Not high-fidelity sine waves by any stretch of the imagination, but good enough for anyone who didn't have "Golden Ears" to easily distinguish the notes. We could have improved it to lower the harmonic distortion created by the triangle-to-sine analog conversion but that wasn't necessary. Probably wouldn't be necessary to drive your cantilever either, at or near resonance.

I think two phase-locked loops generating f1 /4 and f2 /4 from f1 and f2 already summed could be made to work, but it would be difficult to separate f1 from f2 if they are "close together" in frequency. It appears they are not too close together if one is 150 kHz and the other is 160 kHz. OTOH, perhaps I misread what you wrote and one signal is at 150 Hz and it is supposed to amplitude modulate the 160 kHz signal, which appears to be what is happening with the waveform you presented as the cantilever excitation. Could you clarify that?

Just out of curiosity, what physical property causes the cantilever to respond at four times the excitation frequency? Dynamics was not my strongest physics subject.

Hop
 

ehsan66

Mar 3, 2016
6
Joined
Mar 3, 2016
Messages
6
@hevans1944 thank you so much for the detailed reply.

I am using Asylum AFm, a MFP-3D with ARC2 controller. I am sure that ARC2 has two digital lock-ins. We are in contact mode and at the two sides of the resonance (contact resonance freq. between f1 and f2). Unfortunately, f1 and f2 are not independently available on BNC connectors and just the summation is available. f1 and f2 are not very close together (10 kHz difference). I don't think with AM at 160 kHZ, but we simply have the summation of the 150 kHZ and 160 kHz signal. I examined the signal using a specturm analyzer and we have two peaks of same amplitude on 150 and 160 kHz.
On the physics of the problem, ionic movement under the tip contact can cause a fourth harmonic response.

Ehsan
 
Top